Hi @<1523701070390366208:profile|CostlyOstrich36> I ran this with the repo and script arguments, and it seemed that the package installation reverted to cloning the repo.

Executing task id [task]:
repository = my_repo
branch = my_branch
version_num = 
tag = 
docker_cmd = test
entry_point = clearml_pipeline/scheduler.py
working_dir = .

The docker image is correctly identified here, but I still face the installation errors I had before, which I hoped to circumvent using the image

Thanks @<1523701070390366208:profile|CostlyOstrich36> , what would be the intended use case of the docker option?
I was using it because I have some packages I'd like to install from a private repo, with a poetry environment, so I found it easier to containerize this set up process, as some authentication was necessary

I thought the docker option meant I can simply run the task using nothing but the docker image

Yes and yes 🙂

And if so, does it mean the dockerfile isn't necessary?

Thanks @<1523701070390366208:profile|CostlyOstrich36> , so then I must still reference the repo and script?

Hi @<1603560525352931328:profile|BeefyOwl35> , The agent uses it's own entry point, so yes you do need to specify it even if it's in the dockerfile 🙂